Server Manager 8.98.3 keeps hanging

[Michael L.]

List,

I’ve recently gotten involved in a WAS 6.0 to 6.1 on Windows / 8.97.0.2 to 8.98.3.1 TR upgrade. Server Manager, WAS 6.1 and multi-foundation was completed a while back and before I got here. One of the issues we are having is performance in Server Manager to the point SM ‘hangs’ and we need to end java.exe*32, restart the SM service and try logging back in again which can take up to 3 minutes if at all. Navigating for a minute or two in SM hangs SM and we need to repeat this process.

I did find some knowledge (known issue with ES and 64 bit with 2008 on SM) that talked about replacing JDK 1.5 with JDK 1.6 which we have done but has not helped.

Would appreciate it if you could share your thoughts.

Thanks in advance,

8.12, 8.97.0.2 / 8.98.3.1 (multi foundation) iseries 400 ES V5R4 (current w/ group PTF’s, meets MTR’s). Our Deployment server (SM 8.98.3 is installed here) is Windows Server 2008 Enterprise Edition (64-bit) service pack 2 with two 2.66 processors and 8GB of Ram on VMWARES VM. Six Production web servers on 6.1.0.31 with 6 instances on each server (to support 8.98.3.1 but not being used yet) as well as two ES and four more web servers supporting 8.97.0.2.

 

[johncsqd]

I'm not sure if this is your issue (because it was supposed to be fixed in 8.98.2.3), but take a look at this document -

SVM: Server Manager Console Slow Down, Hung, Java Heap Errors. [ID 987184.1]

https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&doctype=PROBLEM&id=987184.1

John

 

[Michael L.]

John, thanks for responding. Like you said this was to be fixed in 8.98.2.3. We'll take a look. 10 days ago our lead CNC was able to add 6 new Production web servers and installed 6 HTML applications servers on each WS via SM w/ no performance issues at all. Last night we did a full PD from the DS and the client finished in just under two hours so ‘the box’ performs well. I’m going to double check the MYTR’s and were checking to see if any windows updates have been automatically applied to the DS. Seems very odd that all of a sudden SM has become unusable due to performance. What I mean by unusable is that if we kill and restart the service it can take 10 minutes to get logged in. Once in we can navigate to another screen but that’s it. Once we try anything else SM ‘hangs”.

Thanks in advance for any ideas.

 

[brother_of_karamazov]

[ QUOTE ]
John, thanks for responding. Like you said this was to be fixed in 8.98.2.3. We'll take a look. 10 days ago our lead CNC was able to add 6 new Production web servers and installed 6 HTML applications servers on each WS via SM w/ no performance issues at all. Last night we did a full PD from the DS and the client finished in just under two hours so ‘the box’ performs well. I’m going to double check the MYTR’s and were checking to see if any windows updates have been automatically applied to the DS. Seems very odd that all of a sudden SM has become unusable due to performance. What I mean by unusable is that if we kill and restart the service it can take 10 minutes to get logged in. Once in we can navigate to another screen but that’s it. Once we try anything else SM ‘hangs”.

Thanks in advance for any ideas.

[/ QUOTE ]

Keep in mind that one can upgrade Server Manager at any time, it does not have to match the release level of the other components. It may be an option to simply install the latest update.

 

[Michael L.]

Jeff, great idea, appreciate the response. I checked and we are 1 update behind. Over the weekend I stopped 42 Production HTML application servers (on 6 WINTEL Web Servers) that are not yet in use (but will be next week) and that has made the performance acceptable, not great. I'll update the thread in the next few days.

 

[brother_of_karamazov]

[ QUOTE ]
Jeff, great idea, appreciate the response. I checked and we are 1 update behind. Over the weekend I stopped 42 Production HTML application servers (on 6 WINTEL Web Servers) that are not yet in use (but will be next week) and that has made the performance acceptable, not great. I'll update the thread in the next few days.

[/ QUOTE ]


"Six Production web servers on 6.1.0.31 with 6 instances on each server"

Didn't really notice this until you mentioned "42 Production HTML application servers (on 6 WINTEL Web Servers)". I am guessing that SM is choking on the large amount of communications with the agents.

How do you have the instances configured? Clustered or individual instances?

 

[Michael L.]

Actually, my original e-mail should have read 7 ports on each server for a total of 42. These 42 (on 8.98.3.1 on new hardware) will replace the existing 42 on 8.97.0.2 this weekend.

We are using the Citrix NetScaler solution to manage our user connections. I'm trying to find out which version.

Thanks in advance for your thoughts and/or concerns.

 

[msouterblight1]

OK, so there are no clusters in use here, Vertically or Horizontally?

 

[johncsqd]

Michael,

Are your WebSphere servers at release 6.1.0.31? I don't see that on the MTR's as being supported yet. I see 6.1.0.27 as the highest service pack. I wonder if that could be part of the problem.

John

 

[Michael L.]

I found some knowledge that talked about "There are no known issues with running the EnterpriseOne server together with the Server Manager agent on 64-bit Windows Server 2008". The interaction of 64-bit Windows Server 2008 and JMX connections causes scfagent to consume an excessive number of Windows "handles". The recommendation was to replace JDK 1.5 with 1.6. This seemed to do the trick.

Thanks to everybody who responded.

Michael